Jan 2022 – present
Remote
4 yrs 9 mos
current
Software Engineer
UpFlux / Process Intelligence
PythonFastAPIPostgreSQLBackground workersPolarsAWS
- Co-designed and continue to evolve the backend architecture of an AI-powered procurement agent, defining API, worker, and service boundaries while owning architectural improvements, performance optimization, and new backend capabilities.
- Re-architected resource-intensive workloads by moving long-running processing from synchronous API requests to background workers, improving resource isolation and service availability.
- Designed and built an internal Python ETL framework for Process Mining applications, standardizing data-processing workflows across CSV, Excel, and Parquet sources and using Polars as the core processing engine.
- Optimized large-scale event transformations by replacing Pandas-based processing with Polars and applying algorithmic improvements, lazy evaluation, parallel execution, and memory-conscious processing techniques, enabling workloads with tens of millions of events.
- Designed and built reusable integration components for enterprise systems, supporting data extraction from SAP, TOTVS/Datasul and Protheus, with resilience mechanisms including retries for transient server failures, rate limiting, and circuit breakers.
- Diagnosed and resolved production memory and performance issues by investigating recurring service restarts, application logs, and RAM metrics, identifying resource-retention problems in caching and document-processing workflows and improving service stability.